save-conversation

Save Conversation

ℹ️ First, set expectations: before anything else, print one short line with this skill's version and its most recent change — the top entry of metadata.changelog above — e.g. save-conversation vX.Y — recent: <summary>. Then proceed.

Capture the current conversation as a markdown file in a standardized folder structure. Unlike automatic session logs, this is selective - only save conversations worth preserving.

---

When to Use

• Completed a meaningful task you want to reference later
• Had a discussion with important decisions or insights
• Debugging session with a solution worth documenting
• Research conversation with findings to preserve
• Any conversation that shouldn't disappear into the void

---

Configuration

Storage Location

Default: saved-conversations/ in the current working directory.

Create a .conversation-config file to customize:

storage_path: ./my-conversations

Or use environment variable: CONVERSATION_STORAGE_PATH

---

Workflow

Step 1: Determine Topic Name

If provided as argument ($ARGUMENTS), use it.

Otherwise, derive from the conversation:

• Identify the primary goal or topic discussed
• Use lowercase-with-hyphens format
• Be descriptive: debug-auth-token-refresh not debugging
• Prefix with category when appropriate: research-, fix-, setup-

If unclear, ask the user for a topic name.

Step 2: Check Storage Location

Check for .conversation-config file:

cat .conversation-config 2>/dev/null || echo "not found"

Check environment variable:

echo "${CONVERSATION_STORAGE_PATH:-not set}"

Default to saved-conversations/ if neither exists.

Step 3: Generate Filename

Format: YYYY-MM-DD_[topic-name].md

Example: 2025-02-20_debug-auth-token-refresh.md

If file already exists, append incrementing number:

• 2025-02-20_topic.md
• 2025-02-20_topic-2.md
• 2025-02-20_topic-3.md

Step 4: Analyze Conversation

Review the current conversation to extract: